黄墩兵,田欢,毋振华,等.老年肌少症患者跌倒影响因素分析及预测模型构建[J].中华物理医学与康复杂志,2026,48(2):148-155
扫码阅读全文
|
| 老年肌少症患者跌倒影响因素分析及预测模型构建 |
|
| |
| DOI：10.3760/cma.j.cn421666-20250603-00459 |
| 中文关键词: 老年人 肌少症 跌倒 影响因素 预测模型 |
| 英文关键词: Elderly Sarcopenia Falls Risk factors Predictive modeling |
| 基金项目:虹口区卫生健康委员会医学科研课题 (虹卫 2302-13)；虹口区第二轮 “国医强优” 三年行动计划 (2022-2024 年) 资助 (HKZYY-2024-11) |
|
| 摘要点击次数: 303 |
| 全文下载次数: 307 |
| 中文摘要: |
| 目的 探讨影响老年肌少症患者跌倒的相关因素，并构建风险预测模型。方法 采用横断面研究方法，收集 2024 年 1 月至 12 月同济大学附属上海市第四人民医院就诊的 318 例老年肌少症患者临床资料，按 7∶3 比例随机分为建模组（223 例）及验证组（95 例）。依据建模组患者近 1 年内是否跌倒分为跌倒组与非跌倒组，进行单因素及多因素 Logistic 回归分析，筛选独立危险因素，构建列线图预测模型。通过 ROC 曲线、Hosmer-Lemeshow 拟合优度检验及临床决策曲线 (DCA) 分析评估模型性能。结果 建模组患者跌倒发生率为 38.12%。单因素分析显示，Berg 平衡量表 (BBS) 评分、闭眼压力中心 (COP) 运动长度、步态周期、步速、股直肌积分肌电值 (iEMG) 及胫骨前肌均方根值 (RMS) 等变量在两组间差异有统计学意义（P<0.05）。多因素回归分析发现，低 BBS 评分、闭眼 COP 运动长度增加、步态周期延长、步速下降、股直肌 iEMG 及胫骨前肌 RMS 降低均为老年肌少症患者跌倒的独立危险因素。构建的列线图模型 ROC 曲线下面积 (AUC) 为 0.837（95% CI：0.785~0.889），灵敏度 74.1%，特异度 83.3%，Hosmer-Lemeshow 检验 χ²=15.5，P=0.061，拟合良好；验证组 AUC 为 0.877（95% CI：0.805~0.950），区分能力良好。结论 整合平衡、步态与表面肌电等多维客观生物力学及神经肌肉指标构建的预测模型，具有良好判别与校准性能，但未纳入 “既往跌倒史” 可能限制模型预测上限及临床适用性，需在多中心人群中进一步验证稳健性及应用边界。 |
| 英文摘要: |
| Objective To identify factors associated with falls among elderly persons with sarcopenia and to develop a risk-prediction model. Methods This cross-sectional study treated clinical data describing 318 elderly hospital patients with sarcopenia. They were randomly divided into a model group (n=223) and a validation group (n=95) and also classified into a fall group and a non-fall group according to their history of falls in the previous year. Univariate analyses and multivariable logistic regression were performed to determine independent predictors of falls. A nomogram was then constructed based on the significant variables, and its performance was evaluated using the receiver operating characteristics (ROC) curve, the Hosmer-Lemeshow goodness-of-fit test, and decision curve analysis. Results In the derivation cohort, the incidence of falls was 38.12%. Univariate analyses showed significant inter-group differences in terms of Berg Balance Scale (BBS) scores, eyes-closed sway path lengths, gait cycle times, gait speed, rectus femoris iEMG, and tibialis anterior RMS. Multivariable logistic regression identified lower BBS scores, longer eyes-closed sway path length, prolonged gait cycle time, slower gait speed, lower rectus femoris iEMG, and lower tibialis anterior RMS as independent predictors of fall risk. The nomogram based on those variables yielded an area under the ROC curve (AUC) of 0.837 (95%CI, 0.785-0.889), with a sensitivity of 74.1% and specificity of 83.3%. The Hosmer-Lemeshow test indicated good calibration (X²=15.5, P=0.061). In the validation cohort, the AUC was 0.877 (95%CI, 0.805-0.950), demonstrating good discrimination ability. Conclusions A fall-risk assessment tool for elderly sarcopenic patients was developed based on objective biomechanical observations rather than prior fall history. The model showed good discrimination and calibration in both derivation and validation cohorts. However, excluding prior fall history, a strong predictor, may limit the model's performance and some aspects of clinical applicability. Therefore, further multicenter studies are warranted to extend and refine the model. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|